Add new comment

Philosophy, Stats, and Pattern Recognition 2

=== Bryned or Rewarded? ===

In the Byrnes POTD, for example... everybody knows what Byrnes' numbers have been.  He was an All-Star in '07 and he was injured the last two years.  We could noodle around all day discussing his RAR in past years, and come up with the same thing everybody else does.

The question is what Byrnes will be in 2010...

The answer being, a very good 3-position platoon OF with a 30% chance of rebounding to his 2003-04 vintage.  As far as we can tell, anyway.

The important thing about Byrnes is that he has SLG'd .500 against LHP's since he fell out of a crib.  He does this while playing center field. 

The secondary question is what his chance is of recapturing the 20-30 runs above RLP with Oakland, or the 40 he had in 2007...

The essential information here includes the knowledge of whether Byrnes trained clean or not, and what Byrnes' historical comps did at age 34.

It may look like philosophy instead of numbers; in reality, it's the attempt to edit noise and identify the numbers that matter.

.

=== Judicious-Lee Used in 2010?, dept. ===

What do you know about Cliff Lee?  Run fifty stats and :shrug: they'll tell you that Lee was great in 2008-09.   What then?  Are we going to discuss whether Cliff Lee is worth $22m vs $26m?  What happens after we decide that?

But, find a pertinent set of comps for Cliff Lee, run their numbers, as below...

Comps list 1

Comps list 2

Comps list 3

...and you might extract some essential information.

In our view, the important stats are the IP columns after the peak years for Lee's comps.  Picture-Perfect Lefties get overused, burnt out, and injured because they look so smooth even as they're about to implode.

.

=== 3-D Information Processing dept. ===

Templates and pattern recognition aren't really within the domain of philosophy.  They're devices used to organize information and statistics so that you deal with fewer, but more important, bits of information.

The goal isn't philosophy over stats; the goal is logic and pattern recognition to weed stats and find the right ones. 

Good ole baseball scouts have been using them for 100 years.  Ken Griffey Jr. came up as the lefthand Willie Mays.  The differences and similarities between the two became, in 1989-1992, the topic of discussion and a good way to predict Junior's career arc.

.....

All information is good, including runs, RBI, and ERA.  Once you have the 5-finger exercises like RAR and FIP down ... can you use them to arrive at the conclusion that reliever Brandon League is a more valuable commodity than starter Brandon Morrow ?  :- )

It's not a question of philosophy-over-stats; it's a 3-dimensional approach to the statistical problem that, in an ideal world, transcends a 2-dimensional $/WAR approach.

.

=== Objection, Your Honor dept. ===

How do you know you've found the essential information?  Who's to say that Aardsma's FB% caused his lower BB rate and that he can't get away with that much longer?

You don't know you've found the essential information.  It's a baseball chat, not NORAD.

The best you can hope for is to propose what might be the essential information, and then let your brainy audience decide for themselves its merit.

Fortunately for us, we're shadow boxing.  Glad we don't face Zduriencik's consequences if he picks the wrong information :- )

Cheers,

Dr D

Interest locations: 

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd><p><br>
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.

shout_filter

  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.
  • Web page addresses and e-mail addresses turn into links automatically.